Design for 600R Bridged-T Pad Attenuator 24 Step Rotary Switch!

[friesdan]

Hey all,
I've been doing quite a few DIY builds in the past year, and just started in on a Drip Fairchild. I'm planning to use it for mastering, and consequently wanted stepped input attenuation.  I noticed that some other compressors and mic pres also use a T-Pad 600R Attenuator, but that you really can't buy a stepped and matched pair of them without spending $500 + (maybe not even?) at goldpoint. So I took it upon myself to design one, and wanted to share the -yet untested - design here, to save anyone else the day's work. Also, maybe some of you pros might find an error? Just posted this on the drip forum, and I'll copy it here:


Here's the Design for the 600R Bridged T Pad Attenuator.
The design uses these very cheap, but clean (just used them on a Diy-racked Meq-5 build) switches:
www.kukaudio.com/au/cpxx.asp?id=94
$5 each for the double-decker ones. Compare that to $159 from Goldpoint!
There are 24 positions on the switches, so I designed the attenuation to match the 0 to -20 dB on the faceplate for the first 21 positions, and then added 3 more "hidden" attenuations of -25, -30, and -40 dB, which won't appear on the faceplate, but will flow naturally as you move the switch counter-clockwise.
The design follows this schematic for a bridged-T:
www.electronics-tutorials.ws/attenuators...ed-t-attenuator.html
What that page calls "R3," I'm calling "R1." My R2 is in the same position.
Deck 1 of the switch is my R1, and Deck 2 is R2.
I used an online calculator to get overall resistance values for the dB attenuation points I wanted (Ideal R1 and Ideal R2) , then put these into excel, and calculated resistor values based on what was available on mouser and came closest. I chose all 2W resistors, mostly metal film, a couple variations.
I'm attaching a pdf of my excel worksheet, with all the values listed.
I'm also attaching a hand-drawn manual for wiring, with resistor values indicated on the switches.
I have never done anything like this before and I'm not 100% sure this will work, so if anyone out there wants to check my work, please do!
Here'a a mouser cart for all the resistors you'll need, plus lots of extras:
www.mouser.com/ProjectManager/ProjectDet...?AccessID=9d71e72714
You might want to edit out the extra resistors if you don't want them, and don't want to pay as much.
Items are labeled as to where they belong on the switches! This took me all day! It's about time I give something back to this community.
If nobody gets up here and corrects my work (It's not guranteed yet!!!) I'll report back after I'm finished.
Total cost: about $130 for 2 of these, with extra resistors to work with for matching channels.
Edit - There was an error in the original schematic/manual, which is now fixed.

 

[friesdan]

Here's the spreadsheet with all the deets.

 

[abbey road d enfer]

The design uses these very cheap, but clean (just used them on a Diy-racked Meq-5 build) switches:
$5 each for the double-decker ones. Compare that to $159 from Goldpoint!

Indeed, they are much cheaper. I can note instantly that the contacts are not gold-plated, which is not a problem for now but may become one in the long run.

I used an online calculator to get overall resistance values for the dB attenuation points I wanted (Ideal R1 and Ideal R2) , then put these into excel, and calculated resistor values based on what was available on mouser and came closest.

I don't want to spend the day checking your calcs (did I mention I don't relish sweating over Excel files?), but it seems like the values are correct.

I chose all 2W resistors,

Why???? There will be only milliwatts in there.

 

[shabtek]

is bridged-tee betta than regular t-pad, gotta check the book

2 poles switch can swap in regular pad with shunt common

anyway nice work, lotta work

you 'll want a reliable switch (as abby implies)

 

[abbey road d enfer]

is bridged-tee betta than regular t-pad, gotta check the book

Bridged-T guarantees more or less constant input AND output impedance. Used to be important in all-valve equipment.

 

[emrr]

Given the labor, it really is worth spending for quality switches, IMO.  I don't want to have to build it again.  Report back about it after 3-5 years. 

1/8 or 1/4W is plenty for this app. 

I've noticed there are multiple approaches to calculating the values in old books, all seem to work in the more limited step versions I've built.

 

[friesdan]

Thanks for the feedback all. Just realized with the help of an engineer friend that there was a mistake in the drawing, now fixed.
Every Drip build I do, there seems to be this attitude of "overbuild everything" that I just go ahead with. I've read so much documentation on the 670 recently that I can't even remember where I read it, but somewhere it said "all resistors should be 2w." Anyway I couldn't see why, but went ahead with it. Thanks for letting me know it isn't necessary.. They also call for all carbon film resistors, which I couldn't manage to make happen either!

 

[friesdan]

Given the labor, it really is worth spending for quality switches, IMO.  I don't want to have to build it again.  Report back about it after 3-5 years. 

I'm only here because I enjoy the labor! I will make a note of reporting back in a few years..

 

[andre tchmil]

Hey all,
I've been doing quite a few DIY builds in the past year, and just started in on a Drip Fairchild. I'm planning to use it for mastering, and consequently wanted stepped input attenuation.  I noticed that some other compressors and mic pres also use a T-Pad 600R Attenuator, but that you really can't buy a stepped and matched pair of them without spending $500 + (maybe not even?) at goldpoint. So I took it upon myself to design one, and wanted to share the -yet untested - design here, to save anyone else the day's work. Also, maybe some of you pros might find an error? Just posted this on the drip forum, and I'll copy it here:


Here's the Design for the 600R Bridged T Pad Attenuator.
The design uses these very cheap, but clean (just used them on a Diy-racked Meq-5 build) switches:
www.kukaudio.com/au/cpxx.asp?id=94
$5 each for the double-decker ones. Compare that to $159 from Goldpoint!
There are 24 positions on the switches, so I designed the attenuation to match the 0 to -20 dB on the faceplate for the first 21 positions, and then added 3 more "hidden" attenuations of -25, -30, and -40 dB, which won't appear on the faceplate, but will flow naturally as you move the switch counter-clockwise.
The design follows this schematic for a bridged-T:
www.electronics-tutorials.ws/attenuators...ed-t-attenuator.html
What that page calls "R3," I'm calling "R1." My R2 is in the same position.
Deck 1 of the switch is my R1, and Deck 2 is R2.
I used an online calculator to get overall resistance values for the dB attenuation points I wanted (Ideal R1 and Ideal R2) , then put these into excel, and calculated resistor values based on what was available on mouser and came closest. I chose all 2W resistors, mostly metal film, a couple variations.
I'm attaching a pdf of my excel worksheet, with all the values listed.
I'm also attaching a hand-drawn manual for wiring, with resistor values indicated on the switches.
I have never done anything like this before and I'm not 100% sure this will work, so if anyone out there wants to check my work, please do!
Here'a a mouser cart for all the resistors you'll need, plus lots of extras:
www.mouser.com/ProjectManager/ProjectDet...?AccessID=9d71e72714
You might want to edit out the extra resistors if you don't want them, and don't want to pay as much.
Items are labeled as to where they belong on the switches! This took me all day! It's about time I give something back to this community.
If nobody gets up here and corrects my work (It's not guranteed yet!!!) I'll report back after I'm finished.
Total cost: about $130 for 2 of these, with extra resistors to work with for matching channels.
Edit - There was an error in the original schematic/manual, which is now fixed.

Looking at the drawing picture, I'm trying to figuring out how this is balanced . Can be that I had too many beers already

 

[abbey road d enfer]

Looking at the drawing picture, I'm trying to figuring out how this is balanced . Can be that I had too many beers already

I would rather think that's because it is not "balanced", although nothing prevents it to be connected in a balanced connection.

 

[abbey road d enfer]

Looking at the drawing picture, I'm trying to figuring out how this is balanced . Can be that I had too many beers already

I would rather think that's because it is not "balanced"; who said it should be?

 

[andre tchmil]

Looking at the drawing picture, I'm trying to figuring out how this is balanced . Can be that I had too many beers already

I would rather think that's because it is not "balanced", although nothing prevents it to be connected in a balanced connection.

in a 4 deck configuration then ? please explain .

 

[emrr]

Don't connect it to ground, connect it between balanced floating inputs.  If you need to run miles and miles you might prefer an H pad (two T's), but it's generally unnecessary. 

 

[abbey road d enfer]

Looking at the drawing picture, I'm trying to figuring out how this is balanced . Can be that I had too many beers already

I would rather think that's because it is not "balanced", although nothing prevents it to be connected in a balanced connection.

in a 4 deck configuration then ? please explain .

 
Indeed, a 4-deck would be needed for a strictly balanced version. But when do you need a balanced attenuator in a balanced connection? As long as the attenuator is floating, it performs its task beautifully. Maybe there's some degradation of HF CMRR due to unbalanced parasitic capacitance, but it has never been a problem with audio signals.

 

[Baltazar]

Would give you a warning on spending time soldering on that switch...
Needed one for a similar project, bought one with pre-mounted resistors (dirt-cheap) and one without (also dirt-cheap). Received both, they look EXACTLY the same but the shafts where a bit different.
Installed it and it jammed after 3 minutes.
I should mention they where not ordered from kukaudio (cock-audio in swedish lol) but they've got to be from the same factory.

Decided I'm not going to spend any time on that switch, soldering resistors, installing etc, so i ordered an Elma instead...

 

[dagoose]

Just wondering, since I need a stepped T pad (stereo, so 4 decks) for the input of my modded PM670.
Is this still the way to go or are there other, easier options?

 

[abbey road d enfer]

Been wanting to try a 2.5K Bridged T

Think the values are close enough and it's correct looking so far. Going off of NYD's drawing obviously .

What are the nominal source and load impedances? Without them, it's impossible to assess.

 

[abbey road d enfer]

The reason for using a bridged-T instead of an L attenuator is that it is constant impedance.
Since you intend to use it with various elements, it's not justified.
You'll have good results with an L attenuator, which is less costly since it uses a single wafer per channel.
Making the taper correct for both 600 and 10k load is impossible but a working compromise is possible.

 

[scott2000]

The reason for using a bridged-T instead of an L attenuator is that it is constant impedance.
Since you intend to use it with various elements, it's not justified

Just trying to visualize and understand, so If a 2.5k Bridged T was used only between a low impedance source (100-200r) dac or other piece and a 20k load? Would that satisfy the 10x bridging thing?
It's also unclear on what the relationship is to the actual attenuation steps as it relates to mismatched impedances. From what I understand, the steps of attenuation won't actually be what is calculated for 2.5k for 1dB since the source and load wouldn't be matched. Is there reading on this? If the first question of the bridging makes it useful in that regard, are the attenuation steps being out of whack the worst thing that will happen? Again, is there reading on this and calculations to find compromises? I feel it's a futile exercise, just curious.

 

[abbey road d enfer]

Just trying to visualize and understand, so If a 2.5k Bridged T was used only between a low impedance source (100-200r) dac or other piece and a 20k load? Would that satisfy the 10x bridging thing?

By tacit convention, bridging is based on the receiver impedance being about 10 times the source impedance, so with a 2.5k bridged-T attenuator driving a 10k input, you don't satisfy this criterion; however, does it matter much? Bridging is essentially about limiting transmission losses, which is not an issue when dealing with...attenuators!

It's also unclear on what the relationship is to the actual attenuation steps as it relates to mismatched impedances. From what I understand, the steps of attenuation won't actually be what is calculated for 2.5k for 1dB since the source and load wouldn't be matched. Is there reading on this? If the first question of the bridging makes it useful in that regard, are the attenuation steps being out of whack the worst thing that will happen? Again, is there reading on this and calculations to find compromises? I feel it's a futile exercise, just curious.

The mathematical answer is very complicated, but fortunately we have simulators.
The result, in brief, is that the first steps are the ones that are the most shifted. E.G. a 2.5k attenuator driven by a 100r source and driving a 10 input sees its first position (which is actually -6) shifted to -0.086dB, the -6dB position shifted to -3.4, its -12 position shifted to -8.6 and the -18 position shifted to -14.5
So the biggest discrepancy is in the first steps, where the 6 db range becomes compressed to -3.4